The stat that commentators love
One of the most overrated research points in Daily Fantasy MLB is Batter vs Pitcher Data (BvP). Normally the more specific a stat, the better, and its hard to get as specific as an individual match up. The problem is sample size. Joey Votto could bat .500 against Clayton Kershaw this year, but it’s obviously nearly impossible for that level of efficiency to continue in the future against Kershaw. This sample wouldn’t even be enough to confidently say Joey Votto’s batting average against Kershaw will be higher than his batting average vs the rest of the league as a whole.
To demonstrate how a players batting average against a pitcher fluctuates over time, here’s Jose Bautista’s batting average over time against Chris Tillman.
Bautista’s first 9 at bats facing Tillman were during the 2009 and 2010 seasons. At that point, with his batting average at .571, you might have assumed Bautista had Tillman’s number. Had you drafted Bautista against Tillman based off this BvP data, you would have had been frustrated as Bautista’s BA plummeted over the course of the next four seasons they faced each other. Excluding the first two seasons, Bautista has only batted .219 against Tillman. The lowest point for his overall BA in this BvP match up was after At Bat 38 where his BA was only .242.
Another issue with BvP data is that batters and pitchers are not always playing at the same level. Over time players typically improve, peak, and decline. Ichiro Suzuki and Bartolo Colon have faced each other 115 times over the course of 2001 to 2015. Ichiro’s hitting abilities have consistently declined, while Bartolo Colon has pitched some of his best baseball in recent years (an anomaly). This makes it reasonable to assume Ichiro’s numbers will be worse against Colon going forward than they have been historically. The point is historical BvP numbers will never demonstrate the ability both players are playing at in the present.
So while it may be tempting to look at batter vs pitcher data for your fantasy lineups, you’re likely to be better off avoiding BvP all together. Stick to stats with some real sample sizes.